Artificial Intelligence in Pancreatic Ductal Adenocarcinoma Imaging: A Commentary on Potential Future Applications
2023; Elsevier BV; Volume: 165; Issue: 2 Linguagem: Inglês
10.1053/j.gastro.2023.04.003
ISSN1528-0012
AutoresMegan Schuurmans, Natália Alves, Pierpaolo Vendittelli, Henkjan Huisman, John J. Hermans, Geert Litjens, David K. Chang, Caroline S. Verbeke, Núria Malats, Matthias Löhr,
Tópico(s)Radiomics and Machine Learning in Medical Imaging
ResumoPancreatic cancer is one of the deadliest cancers worldwide, with a 5-year survival rate of less than 5%.1Bengtsson A. Andersson R. Ansari D. The actual 5-year survivors of pancreatic ductal adenocarcinoma based on real-world data.Sci Rep. 2020; 1016425Crossref Scopus (147) Google Scholar Pancreatic ductal adenocarcinoma (PDAC), the most common and aggressive type of pancreatic cancer, has become a medical emergency in the past decades. PDAC cannot be effectively prevented or screened for and is associated with 98% life expectancy loss and a 30% increase in disability-adjusted life-years.2Michl P. Löhr M. Neoptolemos J.P. et al.UEG position paper on pancreatic cancer. Bringing pancreatic cancer to the 21st century: prevent, detect, and treat the disease earlier and better.United European Gastroenterol J. 2021; 9: 860-871Crossref PubMed Scopus (21) Google Scholar,3Löhr J.M. Pancreatic cancer should be treated as a medical emergency.BMJ. 2014; 349: g5261Crossref PubMed Scopus (24) Google Scholar Still, research funding for PDAC remains significantly lower than for other cancer types, leading it to be flagged as a neglected cancer by both the European Commission and the United States Congress.2Michl P. Löhr M. Neoptolemos J.P. et al.UEG position paper on pancreatic cancer. Bringing pancreatic cancer to the 21st century: prevent, detect, and treat the disease earlier and better.United European Gastroenterol J. 2021; 9: 860-871Crossref PubMed Scopus (21) Google Scholar Cross-sectional imaging, namely computed tomography (CT), magnetic resonance (MR), 18fluoro-2-deoxy-D-glucose positron emission tomography–computed tomography (18FDG PET-CT), and endoscopic ultrasound (EUS), play a crucial role in PDAC management. Nevertheless, current international guidelines for image-based stratification, treatment response prediction, and evaluation are heterogeneous and ineffective.4Elbanna K.Y. Jang H.J. Kim T.K. Imaging diagnosis and staging of pancreatic ductal adenocarcinoma: a comprehensive review.Insights Imaging. 2020; 11: 58Crossref PubMed Scopus (63) Google Scholar Histopathology analysis is considered the criterion standard for PDAC diagnosis and characterization. Still, it remains challenging even for experienced pathologists owing to marked morphologic tumor heterogeneity and the limited amount of tumor tissue in biopsy.5Baxi V. Edwards R. Montalto M. et al.Digital pathology and artificial intelligence in translational medicine and clinical practice.Mod Pathol. 2021; 35: 23-32Abstract Full Text Full Text PDF PubMed Scopus (126) Google Scholar, 6Sántha P. Lenggenhager D. Finstadsveen A. et al.Morphological heterogeneity in pancreatic cancer reflects structural and functional divergence.Cancers (Basel). 2021; 13: 895Crossref PubMed Scopus (12) Google Scholar, 7Fu H. Mi W. Pan B. et al.Automatic pancreatic ductal adenocarcinoma detection in whole slide images using deep convolutional neural networks.Front Oncol. 2021; 11665929Crossref Scopus (19) Google Scholar Moreover, histopathology evaluation of treatment response is imprecise, of limited clinical relevance, and affected by interobserver variation.8Janssen B.V. Tutucu F. van Roessel S. et al.Amsterdam International Consensus Meeting: tumor response scoring in the pathology assessment of resected pancreatic cancer after neoadjuvant therapy.Mod Pathol. 2021; 34: 4-12Abstract Full Text Full Text PDF PubMed Scopus (30) Google Scholar Artificial intelligence (AI) has gained considerable interest in oncology because it has the potential to leverage high amounts of data to produce individualized recommendations based on each patient's clinical picture.9Kann B.H. Hosny A. Aerts H.J.W.L. Artificial intelligence for clinical oncology.Cancer Cell. 2021; 39: 916-927Abstract Full Text Full Text PDF PubMed Scopus (108) Google Scholar As the volume of multi-modal data acquired in routine clinical practice increases, AI can support clinicians and ultimately guide decision making at each step of the patient pathway by focusing on well validated applications at meaningful clinical touch-points.9Kann B.H. Hosny A. Aerts H.J.W.L. Artificial intelligence for clinical oncology.Cancer Cell. 2021; 39: 916-927Abstract Full Text Full Text PDF PubMed Scopus (108) Google Scholar Commercial clinical AI is already a reality for diseases like breast and lung cancer, with multiple FDA-approved products on the market for screening, diagnosis, and tumor characterization.10Leeuwen K.G. van Schalekamp S. Rutten M.J.C.M. et al.Artificial intelligence in radiology: 100 commercially available products and their scientific evidence.Eur Radiol. 2021; 31: 3797-3804Crossref PubMed Scopus (140) Google Scholar Currently, there are 2 main approaches for image-based AI: radiomics and convolutional neural networks (CNNs). Radiomics predicts an outcome by feeding manually defined texture and shape features extracted from a region of interest to machine-learning models. CNNs, on the other hand, automatically compute the relevant features directly from the imaging during training, in a neural network comprising a sequence of convolutional and pooling operations. Since the introduction of AlexNet in 2012, CNNs have evolved enormously and are now dominating image analysis, but the transition from hand-crafted radiomic features to deep learning in the medical domain has been gradual.11Krizhevsky A. Sutskever I. Hinton G.E. ImageNet classification with deep convolutional neural networks.Commun ACM. 2017; 60: 84-90Crossref Scopus (16281) Google Scholar,12Litjens G. Kooi T. Bejnordi B.E. et al.A survey on deep learning in medical image analysis.Med Image Anal. 2017; 42: 60-88Crossref PubMed Scopus (7742) Google Scholar The number of publications on AI for clinical decision making in oncology has increased exponentially in the past few years.11Krizhevsky A. Sutskever I. Hinton G.E. ImageNet classification with deep convolutional neural networks.Commun ACM. 2017; 60: 84-90Crossref Scopus (16281) Google Scholar However, AI research in PDAC is still at a preliminary stage compared with other cancer diseases, with limited private and public data sets and a lack of independent external model validation.13Schuurmans M. Alves N. Vendittelli P. et al.Setting the research agenda for clinical artificial intelligence in pancreatic adenocarcinoma imaging.Cancers (Basel). 2022; 14: 3498Crossref PubMed Scopus (2) Google Scholar As a result, no AI applications have been implemented in clinical practice for PDAC. The first step toward clinically relevant AI is to define the research questions to be addressed by AI algorithms. This should be done based on specific patient pathways, by identifying the critical touch-points that are lacking in clinical practice and where AI could have the greatest impact. For this commentary, an international, multi-disciplinary, multi-institutional expert panel including AI experts, pancreatic radiologists, pathologists, and surgeons came together to define the PDAC patient clinical pathway and derive its main touch-points for AI development.14PANCAIM A European consortium to improve pancreatic cancer treatment with artificial intelligence optimizing and integrating genomics and medical imaging.https://pancaim.euDate accessed: September 21, 2022Google Scholar The expert board divided the patient pathway into 5 steps: detection, diagnosis, staging, treatment, and monitoring, as depicted in Figure 1. In each step, the most relevant patient and clinician decision-oriented touch-points for image-based AI research were identified. These touch-points regard clinical decisions that are suboptimal with currently implemented workflows and guidelines and are detailed in the subsequent sections.9Kann B.H. Hosny A. Aerts H.J.W.L. Artificial intelligence for clinical oncology.Cancer Cell. 2021; 39: 916-927Abstract Full Text Full Text PDF PubMed Scopus (108) Google Scholar Timely detection is crucial to improve PDAC patients' outcomes, because the 5-year survival increases from only 3% in metastatic patients to 42% when the tumor is still confined to the primary site.15American Cancer SocietySurvival rates for pancreatic cancer.https://www.cancer.org/cancer/pancreatic-cancer/detection-diagnosis-staging/survival-rates.htmlDate accessed: April 6, 2022Google Scholar According to the Japan Pancreatic Cancer Registry, patients in the earliest disease stage show a survival rate as high as 80.4% but account for only 0.8% of cases.16Egawa S. Toma H. Ohigashi H. et al.Japan pancreatic cancer registry; 30th year anniversary: Japan Pancreas Society.Pancreas. 2012; 41: 985-992Crossref PubMed Scopus (301) Google Scholar Screening groups at risk for PDAC is still cost-prohibitive owing to the relatively low incidence and the absence of validated noninvasive tumor biomarkers. The most used modality for PDAC detection is multi-phase contrast-enhanced CT (CECT). However, early PDAC detection on CECT remains challenging, because lesions are small (<2 cm), present poorly defined margins, and are more often iso-attenuating.4Elbanna K.Y. Jang H.J. Kim T.K. Imaging diagnosis and staging of pancreatic ductal adenocarcinoma: a comprehensive review.Insights Imaging. 2020; 11: 58Crossref PubMed Scopus (63) Google Scholar,17Yoon S.H. Lee J.M. Cho J.Y. et al.Small (≤20 mm) pancreatic adenocarcinomas: analysis of enhancement patterns and secondary signs with multiphasic multidetector CT.Radiology. 2011; 259: 442-452Crossref PubMed Scopus (197) Google Scholar Radiologists' sensitivity at detecting lesions with size smaller than 2 cm on CECT has been reported to be as low as 58%.4Elbanna K.Y. Jang H.J. Kim T.K. Imaging diagnosis and staging of pancreatic ductal adenocarcinoma: a comprehensive review.Insights Imaging. 2020; 11: 58Crossref PubMed Scopus (63) Google Scholar,17Yoon S.H. Lee J.M. Cho J.Y. et al.Small (≤20 mm) pancreatic adenocarcinomas: analysis of enhancement patterns and secondary signs with multiphasic multidetector CT.Radiology. 2011; 259: 442-452Crossref PubMed Scopus (197) Google Scholar Contrast-enhanced MRI is highly effective at detecting tumors that are poorly visible on CECT, but it is not yet routinely implemented in the clinic.18Kim J.H. Park S.H. Yu E.S. et al.Visually isoattenuating pancreatic adenocarcinoma at dynamic-enhanced CT: frequency, clinical and pathologic characteristics, and diagnosis at imaging examinations.Radiology. 2010; 257: 87-96Crossref PubMed Scopus (189) Google Scholar Early detection, arguably the most pressing issue in PDAC management, can be facilitated by the timely identification of secondary imaging signs predictive of PDAC, such as main pancreatic duct cutoff or dilation, parenchymal atrophy, and irregular pancreatic contour.4Elbanna K.Y. Jang H.J. Kim T.K. Imaging diagnosis and staging of pancreatic ductal adenocarcinoma: a comprehensive review.Insights Imaging. 2020; 11: 58Crossref PubMed Scopus (63) Google Scholar,19Singh D.P. Sheedy S. Goenka A.H. et al.Computerized tomography scan in pre-diagnostic pancreatic ductal adenocarcinoma: Stages of progression and potential benefits of early intervention: a retrospective study.Pancreatology. 2020; 20: 1495-1501Crossref PubMed Scopus (29) Google Scholar These signs are often visible on CECT scans 18 to 12 months before clinical diagnosis, but the reported radiologists' sensitivity for their timely detection is only 44%, limiting the chances of early action.19Singh D.P. Sheedy S. Goenka A.H. et al.Computerized tomography scan in pre-diagnostic pancreatic ductal adenocarcinoma: Stages of progression and potential benefits of early intervention: a retrospective study.Pancreatology. 2020; 20: 1495-1501Crossref PubMed Scopus (29) Google Scholar Recent articles have shown great potential for AI-driven PDAC diagnosis in CT.13Schuurmans M. Alves N. Vendittelli P. et al.Setting the research agenda for clinical artificial intelligence in pancreatic adenocarcinoma imaging.Cancers (Basel). 2022; 14: 3498Crossref PubMed Scopus (2) Google Scholar Chen et al20Chen P.T. Wu T. Wang P. et al.Pancreatic cancer detection on CT scans with deep learning: a nationwide population-based study.Radiology. 2023; 306: 172-182Crossref PubMed Scopus (29) Google Scholar developed an algorithm to distinguish between pancreatic cancer and normal pancreas on portal venous CT, which was trained with a large data set of 2011 cases (546 pancreatic cancer). In a test set of 1473 CT studies (669 malignant) from institutions throughout Taiwan, AI achieved an area under the receiver operating characteristic curve (AUC) of 0.95 (95% confidence interval [CI] 0.94–0.96), with 74.7% (68 of 91, 95% CI 64.5–83.3) sensitivity for malignancies smaller than 2 cm. In the internal test set, AI achieved an AUC of 0.96 (95% CI 0.94–0.99), without a significant difference in sensitivity compared with the original radiologists' report. However, the data sets used for training and testing the algorithms are not consecutive but artificially curated, with control cases being derived from liver and renal donors. In practice, patients with suspicion of PDAC often show one or several pancreatic alterations, and to be clinically relevant AI should be able to distinguish PDAC from other less aggressive pancreatic neoplasms. Training AI models with such artificially curated cohorts could cause performance overestimation. Regarding early detection, Mukherjee et al21Mukherjee S. Patra A. Khasawneh H. et al.Radiomics-based machine-learning models can detect pancreatic cancer on prediagnostic computed tomography scans at a substantial lead time before clinical diagnosis.Gastroenterology. 2022; 163: 1435-1446.e3Abstract Full Text Full Text PDF PubMed Scopus (40) Google Scholar studied whether a radiomics-based AI algorithm could detect PDAC at the prediagnostic stage (3–36 months before clinical diagnosis). The study included 155 patients and an age-matched cohort of 265 subjects with normal pancreas for model development and an independent internal set of 176 patients and 80 publicly available control cases for testing.21Mukherjee S. Patra A. Khasawneh H. et al.Radiomics-based machine-learning models can detect pancreatic cancer on prediagnostic computed tomography scans at a substantial lead time before clinical diagnosis.Gastroenterology. 2022; 163: 1435-1446.e3Abstract Full Text Full Text PDF PubMed Scopus (40) Google Scholar The model achieved a high AUC of 0.98 (95% CI 0.94–0.98), significantly outperforming 2 radiologists who independently reviewed images in the test set (mean AUC 0.66, 95% CI 0.46–0.86).21Mukherjee S. Patra A. Khasawneh H. et al.Radiomics-based machine-learning models can detect pancreatic cancer on prediagnostic computed tomography scans at a substantial lead time before clinical diagnosis.Gastroenterology. 2022; 163: 1435-1446.e3Abstract Full Text Full Text PDF PubMed Scopus (40) Google Scholar Despite these promising early results, the identification of small lesions and secondary anatomic signs is still widely disregarded in AI-based detection research, and most studies do not disaggregate performance based on tumor size and stage.13Schuurmans M. Alves N. Vendittelli P. et al.Setting the research agenda for clinical artificial intelligence in pancreatic adenocarcinoma imaging.Cancers (Basel). 2022; 14: 3498Crossref PubMed Scopus (2) Google Scholar In addition, there is a lack of research on lesion localization and a general absence of well curated data sets, with positive and negative cases being retrieved from completely different populations, which does not reflect the clinical landscape and can introduce bias.13Schuurmans M. Alves N. Vendittelli P. et al.Setting the research agenda for clinical artificial intelligence in pancreatic adenocarcinoma imaging.Cancers (Basel). 2022; 14: 3498Crossref PubMed Scopus (2) Google Scholar For AI to improve PDAC detection, it is crucial to acquire and make publicly available well curated, multi-modal data sets that contain a significant proportion of small (<2 cm or even <1 cm) tumors, which should be treated as a subgroup of interest when reporting model performance. PDAC symptoms are mostly non-specific in early disease stages, and because lesional appearances are heterogeneous on CECT, patients are often initially misdiagnosed with other, more common abdominal diseases with similar symptomatology (eg, gallbladder diseases, acute or chronic pancreatitis, duodenum cancer).18Kim J.H. Park S.H. Yu E.S. et al.Visually isoattenuating pancreatic adenocarcinoma at dynamic-enhanced CT: frequency, clinical and pathologic characteristics, and diagnosis at imaging examinations.Radiology. 2010; 257: 87-96Crossref PubMed Scopus (189) Google Scholar,19Singh D.P. Sheedy S. Goenka A.H. et al.Computerized tomography scan in pre-diagnostic pancreatic ductal adenocarcinoma: Stages of progression and potential benefits of early intervention: a retrospective study.Pancreatology. 2020; 20: 1495-1501Crossref PubMed Scopus (29) Google Scholar Initially misdiagnosed patients are reported to present higher rates of abdominal pain, weight loss, and acute pancreatitis than correctly diagnosed patients and are at a higher risk of advanced disease.18Kim J.H. Park S.H. Yu E.S. et al.Visually isoattenuating pancreatic adenocarcinoma at dynamic-enhanced CT: frequency, clinical and pathologic characteristics, and diagnosis at imaging examinations.Radiology. 2010; 257: 87-96Crossref PubMed Scopus (189) Google Scholar Histopathology assessment is the current criterion standard for PDAC diagnosis confirmation and is usually based on EUS fine-needle cytology or biopsy. Nevertheless, the morphologic distinction of PDAC from other lesions on small biopsies or cytology samples can be challenging, especially given the minimal amount of lesional material that is often contained in these samples.19Singh D.P. Sheedy S. Goenka A.H. et al.Computerized tomography scan in pre-diagnostic pancreatic ductal adenocarcinoma: Stages of progression and potential benefits of early intervention: a retrospective study.Pancreatology. 2020; 20: 1495-1501Crossref PubMed Scopus (29) Google Scholar CT with or without contrast is the main modality for PDAC detection. Park et al22Park H.J. Shin K. You M.W. et al.Deep learning-based detection of solid and cystic pancreatic neoplasms at contrast-enhanced CT.Radiology. 2023; 306: 140-149Crossref PubMed Scopus (14) Google Scholar developed a deep-learning model to differentiate images with pancreatic neoplasms (PDAC, neuroendocrine neoplasm, solid pseudopapillary neoplasm, intraductal pancreatic mucinous neoplasm, serous cystic neoplasm, and mucinous cystic neoplasm) from images without pancreatic abnormalities. The authors trained the model in a data set of 852 patients (503 pancreatic neoplasms) and tested it in 2 neoplasm-enriched consecutive data sets (one internal and one external) of patient undergoing contrast CT scans. Furthermore, 2 board-certified radiologists independently interpreted the CT images of the test sets. In the internal test set AI achieved an AUC of 0.91 (95% CI 0.89–0.94), showing no statistically significant difference from the radiologists' performance. However, radiologists performed significantly better than AI in the external test set (AI: AUC 0.87, 95% CI 0.84–0.89; radiologist 1: AUC 0.95, 95% CI 0.93–0.97; radiologist 2: AUC 0.96, 95% CI 0.94–0.97). Current research separates detection, defined as the distinction between PDAC patients and healthy control subjects, from differential diagnosis, defined as the distinction between PDAC and other types of pancreatic lesions.13Schuurmans M. Alves N. Vendittelli P. et al.Setting the research agenda for clinical artificial intelligence in pancreatic adenocarcinoma imaging.Cancers (Basel). 2022; 14: 3498Crossref PubMed Scopus (2) Google Scholar The previously described studies indicate that AI trained with large data sets can approach expert-level performance.22Park H.J. Shin K. You M.W. et al.Deep learning-based detection of solid and cystic pancreatic neoplasms at contrast-enhanced CT.Radiology. 2023; 306: 140-149Crossref PubMed Scopus (14) Google Scholar,23Chen P.-T. Chang D. Yen H. et al.Radiomic features at CT can distinguish pancreatic cancer from noncancerous pancreas.Radiol Imaging Cancer. 2021; 3e210010Crossref Scopus (15) Google Scholar However, both studies focus on binary classification as opposed to differential diagnosis, and the evidence for radiologists' performance is limited because no multi-institutional reader studies have been conducted. It is crucial to move toward well curated data sets including a panoply of relevant pancreatic alterations that should be distinguishable from PDAC. In the future, research should strive toward a single use case for radiology-based AI in PDAC diagnosis that includes both the detection of a lesion and its correct classification among a variety of pancreatic diseases. The current priority is the curation of large data sets with representative percentages of each lesion type and the integration of different imaging modalities that offer complementary information regarding lesion characterization. Research in AI for histopathologic PDAC diagnosis is scarce.13Schuurmans M. Alves N. Vendittelli P. et al.Setting the research agenda for clinical artificial intelligence in pancreatic adenocarcinoma imaging.Cancers (Basel). 2022; 14: 3498Crossref PubMed Scopus (2) Google Scholar Although histopathology is considered the criterion standard for confirming PDAC diagnosis, it is a time-consuming process that suffers from non-uniform implementation in clinical practice and inter-observer variability.13Schuurmans M. Alves N. Vendittelli P. et al.Setting the research agenda for clinical artificial intelligence in pancreatic adenocarcinoma imaging.Cancers (Basel). 2022; 14: 3498Crossref PubMed Scopus (2) Google Scholar Developing powerful AI models for histopathologic PDAC diagnosis is fundamental to advancing AI research at all steps of the patient pathway. Such models would optimize clinical workflows and empower the generation of reliable ground truth that could be used to develop AI with other (noninvasive) modalities in a timely and cost-effective manner. After histopathology diagnosis, the most used method for PDAC staging is the TNM classification by the American Joint Committee on Cancer (AJCC). The local tumor extent (T stage), the dissemination to the regional lymph nodes (N stage), and the metastatic spread to distant sites (M stage) are used to stratify patients, determine their prognosis, and indicate treatment and monitoring strategy.24Edge S.B. Compton C.C. The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM.Ann Surg Oncol. 2010; 17: 1471-1474Crossref PubMed Scopus (6740) Google Scholar Nevertheless, the TNM classification's predictiveness for overall survival (OS) is not reliable.25Song W. Miao D.L. Chen L. Nomogram for predicting survival in patients with pancreatic cancer.Onco Targets Ther. 2018; 11: 539-545Crossref PubMed Scopus (51) Google Scholar A 2018 multi-center study aiming to validate the 8th-edition AJCC TNM in a cohort of 1525 patients receiving pancreatoduodenectomy reported a concordance index of 0.57 (95% CI 0.55–0.60) for OS prediction.26van Roessel S. Kasumova G.G. Verheij J. et al.International validation of the eighth edition of the American Joint Committee on Cancer (AJCC) TNM staging system in patients with resected pancreatic cancer.JAMA Surg. 2018; 153e183617Crossref PubMed Scopus (182) Google Scholar AI for PDAC staging lacks a solid reference standard.13Schuurmans M. Alves N. Vendittelli P. et al.Setting the research agenda for clinical artificial intelligence in pancreatic adenocarcinoma imaging.Cancers (Basel). 2022; 14: 3498Crossref PubMed Scopus (2) Google Scholar TNM staging and histopathologic grade do not correlate sufficiently with OS and suffer from inter-reader variability.13Schuurmans M. Alves N. Vendittelli P. et al.Setting the research agenda for clinical artificial intelligence in pancreatic adenocarcinoma imaging.Cancers (Basel). 2022; 14: 3498Crossref PubMed Scopus (2) Google Scholar A recent systematic literature review identified 13 publications on AI for PDAC staging, of which only 1 considered OS as the ground truth.13Schuurmans M. Alves N. Vendittelli P. et al.Setting the research agenda for clinical artificial intelligence in pancreatic adenocarcinoma imaging.Cancers (Basel). 2022; 14: 3498Crossref PubMed Scopus (2) Google Scholar A study by Chaddad et al27Chaddad A. Sargos P. Desrosiers C. Modeling texture in deep 3D CNN for survival analysis.IEEE J Biomed Health Inform. 2021; 25: 2454-2462Crossref PubMed Scopus (15) Google Scholar divided patients into short- and long-term survivors with a set threshold, achieving 0.72 AUC on an internal testing set, but no external evaluation was performed. In the absence of an international consensus that relates surrogate end points, AI research using clinically obtained low- and high-grade differentiation and predictive TNM is not clinically relevant. Future AI research should focus on discovering new data-driven staging biomarkers that relate histopathology and imaging to OS. The most common treatment options for PDAC are resection and chemo(radio)therapy, in particular with the use of FOLFIRINOX and gemcitabine-abraxane.2Michl P. Löhr M. Neoptolemos J.P. et al.UEG position paper on pancreatic cancer. Bringing pancreatic cancer to the 21st century: prevent, detect, and treat the disease earlier and better.United European Gastroenterol J. 2021; 9: 860-871Crossref PubMed Scopus (21) Google Scholar Surgical resection (Rx) is the only option for potential long-term survival, but as shown in Figure 1 it is suitable for only a minority (10%–15%) of patients (stages I and II). Most patients are diagnosed in later disease stages (III and IV), where Rx is no longer possible owing to metastasis or extensive vessel involvement.28Wittel U.A. Lubgan D. Ghadimi M. et al.Consensus in determining the resectability of locally progressed pancreatic ductal adenocarcinoma—results of the Conko-007 multicenter trial.BMC Cancer. 2019; 19: 979Crossref PubMed Scopus (23) Google Scholar Imaging assessment of tumor-vascular contact primarily determines eligibility for Rx, but there are no widely accepted evidence-based guidelines for appropriate tumor resectability criteria.4Elbanna K.Y. Jang H.J. Kim T.K. Imaging diagnosis and staging of pancreatic ductal adenocarcinoma: a comprehensive review.Insights Imaging. 2020; 11: 58Crossref PubMed Scopus (63) Google Scholar,29Hong S.B. Lee S.S. Kim J.H. et al.Pancreatic cancer CT: prediction of resectability according to NCCN criteria.Radiology. 2018; 289: 710-718Crossref PubMed Scopus (66) Google Scholar As a result, the 5-year survival rate of resected PDAC patients is only 30% to 58%, with 69% to 75% of patients relapsing within 2 years.1Bengtsson A. Andersson R. Ansari D. The actual 5-year survivors of pancreatic ductal adenocarcinoma based on real-world data.Sci Rep. 2020; 1016425Crossref Scopus (147) Google Scholar,30Lambert A. Schwarz L. Borbath I. et al.An update on treatment options for pancreatic adenocarcinoma.Ther Adv Med Oncol. 2019; 111758835919875568Crossref PubMed Scopus (136) Google Scholar As illustrated in Figure 1, most patients receive chemo(radio)therapy at some point during treatment.31Latenstein A.E.J. van der Geest L.G.M. Bonsing B.A. et al.Nationwide trends in incidence, treatment and survival of pancreatic ductal adenocarcinoma.Eur J Cancer. 2020; 125: 83-93Abstract Full Text Full Text PDF PubMed Scopus (88) Google Scholar Neo-adjuvant chemo(radio)therapy (nCTx) intends to optimize surgical outcome in patients with resectable disease, and adjuvant chemo(radio)therapy (aCTx) is used to down-stage non-resectable patients. After aCTx, patients may become resectable and undergo Rx or be referred to palliative care (Px), which is intended to suppress disease-related pain and lengthen the patient's life. Although most patients experience chemotherapy-induced toxicity, often with limited efficacy due to biological resistance, a priori prediction of chemotherapy response is still not possible in current clinical work-up.32Pearce A. Haas M. Viney R. et al.Incidence and severity of self-reported chemotherapy side effects in routine care: a prospective cohort study.PLoS One. 2017; 12e0184360Crossref Scopus (303) Google Scholar,33Harder F.N. Jungmann F. Kaissis G.A. et al.[18F]FDG PET/MRI enables early chemotherapy response prediction in pancreatic ductal adenocarcinoma.EJNMMI Res. 2021; 11: 70Crossref PubMed Scopus (9) Google Scholar Treatment response prediction with the use of AI is a challenging task. Healy et al34Healy G.M. Salinas-Miranda E. Jain R. et al.Pre-operative radiomics model for prognostication in resectable pancreatic adenocarcinoma with external validation.Eur Radiol. 2022; 32: 2492-2505Crossref PubMed Scopus (17) Google Scholar performed a retrospective, international, multi-center study for prognostication in resectable cases with the use of radiomics of pre-operative CT scans combined with clinical factors. The training cohort included 352 patients from 5 Canadian hospitals, and the model was tested on an external set of 215 from 34 hospitals in Ireland. The clinical-radiomic model discrimination (c-index 0.545, 95% CI 0.543–0.546) was higher than TNM (c-index 0.525, 95% CI 0.524–0.526), with P < .001. Despite superiority to TNM, the low model discrimination results in limited clinical utility for potential treatment decisions. Another study, by Yao et al,35Yao J. Shi Y. Cao K. et al.DeepPrognosis: preoperative prediction of pancreatic cancer survival and surgical margin via comprehensive understanding of dynamic contrast-enhanced CT imaging and tumor-vascular contact parsing.Med Image Anal. 2021; 73102150Abstract Full Text Full Text PDF Scopus (23) G
Referência(s)